fix

core/trainers/single_trainer.py

@@ -132,7 +132,7 @@ class SingleTrainer(TranspileTrainer):
         padding = 0
 
         if type_name == "DataLoader":
-            return None#self._get_dataloader(dataset_name)
+            return None
         else:
             return self._get_dataset(dataset_name)
 
@@ -243,7 +243,6 @@ class SingleTrainer(TranspileTrainer):
         metrics_varnames = []
         metrics_format = []
         fetch_period = 20
-        #metrics_format.append("{}: {{}}".format("epoch"))
         metrics_format.append("{}: {{}}".format("batch"))
         for name, var in model_class.get_metrics().items():
             metrics_varnames.append(var.name)
@@ -259,7 +258,7 @@ class SingleTrainer(TranspileTrainer):
                 while True:
                     metrics_rets = self._exe.run(program=program,
                                                  fetch_list=metrics_varnames)
-                    metrics = [batch_id]#[epoch, batch_id]
+                    metrics = [batch_id]
                     metrics.extend(metrics_rets)
 
                     if batch_id % fetch_period == 0 and batch_id != 0:
@@ -275,7 +274,6 @@ class SingleTrainer(TranspileTrainer):
         dirname = envs.get_global_env("epoch.init_model_path", None)
         if dirname is None:
             return
-        dirname = os.path.join(dirname, str(epoch_id))
         if is_fleet:
             fleet.load_persistables(self._exe, dirname)
         else:

models/rank/dnn/config.yaml

@@ -12,43 +12,52 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-debug: false
-cold_start: true
+# 轮数
 epochs: 10
+# 设备
 device: cpu
+# 工作目录
 workspace: "paddlerec.models.rank.dnn"
 
+# dataset列表
 dataset:
-- name: dataset_2
+- name: dataset_2 # 名字，用来区分不同的dataset
   batch_size: 2
-  #type: QueueDataset
-  type: DataLoader
-  data_path: "{workspace}/data/sample_data/train"
+  type: DataLoader # 或者QueueDataset 
+  data_path: "{workspace}/data/sample_data/train" # 数据路径
   sparse_slots: "click 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26"
   dense_slots: "dense_var:13"
 
+# 超参数
 hyper_parameters:
+  #优化器
   optimizer:
     class: Adam
     learning_rate: 0.001
     strategy: async
+  # 用户自定义
   sparse_inputs_slots: 27
   sparse_feature_number: 1000001
   sparse_feature_dim: 9
   dense_input_dim: 13
   fc_sizes: [512, 256, 128, 32]
 
+# executor配置
 epoch:
   name:
   trainer_class: single
-  save_checkpoint_interval: 2
-  save_inference_interval: 4
-  save_checkpoint_path: "increment"
-  save_inference_path: "inference"
+  save_checkpoint_interval: 2 # 保存模型
+  save_inference_interval: 4 # 保存预测模型
+  save_checkpoint_path: "increment" # 保存模型路径
+  save_inference_path: "inference" # 保存预测模型路径
+  #save_inference_feed_varnames: [] # 预测模型feed vars
+  #save_inference_fetch_varnames: [] # 预测模型 fetch vars
+  #init_model_path: "xxxx" # 加载模型
 
+# 执行器，每轮要跑的所有模型
 executor:
  - name: train
-   model: "{workspace}/model.py"
-   dataset_name: dataset_2
-   thread_num: 1
-   is_infer: False
+   model: "{workspace}/model.py" # 模型路径
+   dataset_name: dataset_2 # 名字，用来区分不同的阶段
+   thread_num: 1 # 线程数
+   is_infer: False # 是否是infer